CN110570345B - Blind watermarking method for airspace color digital image fused with discrete cosine transform - Google Patents

Abstract

The invention discloses a blind watermarking method of an airspace color digital image fused with discrete cosine transform by combining the advantages of high running speed of an airspace digital watermarking algorithm and high robustness of a frequency domain digital watermarking algorithm. According to the unique characteristics of the discrete cosine transform direct current coefficients, the invention utilizes the correlation principle of the direct current coefficients of adjacent pixel blocks to finish the embedding and blind extraction of the digital watermark in a space domain without carrying out real discrete cosine transform. The invention embeds the color digital watermark image into the color carrier image, has stronger robustness and higher real-time performance on the premise of better watermark invisibility, solves the problems of low running speed and weak robustness of the digital watermark algorithm of the large-capacity color image, and is suitable for the situation of fast and high-efficiency digital media copyright protection.

Description

Blind watermarking method for airspace color digital image fused with discrete cosine transform

Technical Field

The invention belongs to the technical field of network space security, and relates to copyright protection of a color digital image as a digital watermark.

Background

With the rapid development of multimedia technology, efficient storage and transmission of digitized information is possible, but at the same time, a series of serious problems such as piracy, infringement, tampering, etc. are also generated. Therefore, copyright protection is becoming a widespread concern for students at home and abroad. The digital watermark technology is an effective digital product copyright protection and data security maintenance technology, and the successful embedding and extraction of the watermark can effectively protect the digital copyright, thereby solving the problem of copyright protection.

According to different processing modes of carrier images, the digital watermarking algorithm is divided into a spatial watermarking algorithm and a frequency domain watermarking algorithm, wherein the spatial watermarking algorithm has the advantages of simplicity and rapidness in operation, but the robustness is poor, and the frequency domain watermarking algorithm has the advantage of strong robustness, but the forward transformation and the reverse transformation of the corresponding transformation domain enable the operation time of the watermarking algorithm to be longer. Therefore, how to combine the advantages of the two, and on the premise of ensuring good invisibility of the digital watermark, designing a digital watermark algorithm with strong robustness and high instantaneity becomes a current urgent problem to be solved.

Disclosure of Invention

The invention aims to provide a blind watermarking method of an airspace color digital image fused with discrete cosine transform, which comprises two specific watermarking embedding and watermarking extracting processes, wherein the watermarking embedding process is described as follows:

the first step: one width is of the sizeM×MIs a color carrier image of (a)HThree layered carrier images separated into red, green and blueH _i And imaging each layered carrierH _i Is divided intom×mNon-overlapping pixel blocks of (1), whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a second step of: one width is of the sizeN×NColor watermark image of (a)WDividing into three layered watermark images of red, green and blue, and simultaneously, carrying out key-based on each layered watermark image in order to improve the security of the watermarkKa _i Arnold transformation to obtain three scrambled layered watermark imagesW _i The method comprises the steps of carrying out a first treatment on the surface of the Layered watermark imageW _i Each decimal pixel value in (2) is converted into 8-bit binary number which is sequentially connected into a length of 8N ² Watermark bit sequence of (a)SW _i Whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a third step of: sequentially from layered carrier images in longitudinal orderH _i Is selected adjacent pixel square blockAAndBwhereini=1, 2, 3 represent red, green, blue three layers, respectively;

fourth step: according to formula (1), pixel blocks are directly calculated in the space domainAAndBdirect current coefficient of (2)dc _p ；

(1)

Wherein,,p=1, 2 respectively represents pixel blocksAAndB，mis the size of the row (column) of the pixel block,f _p (x, y) Is a pixel square blockpFirst, thexLine 1yPixel values of the columns;

fifth step: sequentially from a layered watermark sequenceSW _i Selecting watermark bits to be embedded inwThe method comprises the steps of carrying out a first treatment on the surface of the Embedding watermark information according to the correlation of DC coefficients of adjacent pixel blocks by using formulas (2) and (3)wObtaining a pixel blockpModified DC coefficientdc _p ^* ，p=1, 2 respectively represents pixel blocksAAndB；

(2)

(3)

wherein,,win order for the watermark bits to be embedded,avg=(dc ₁ +dc ₂ )/2，T _i is the firstiThe quantization step size of the layer is determined,i=1, 2, 3 represent red, green, blue three layers, respectively;

sixth step: according to equation (4), using the modified pixel valuef _p ^* (x, y) Replacement of original pixel blockpPixel value of corresponding positionf _p (x, y) Obtaining pixel square block containing watermarkA ^* AndB ^* whereinp=1, 2；

(4)

Seventh step: print pixel blockA ^* AndB ^* respectively updated to its layered carrier imageH _i Corresponding positions in (a), whereini=1, 2, 3, respectively representing three layers of red, green, blue;

eighth step: repeating the third to seventh steps until all watermark information is embedded, thereby obtaining layered carrier image containing watermarkH _i ^* The method comprises the steps of carrying out a first treatment on the surface of the Finally, three layers of the composite materialLayered carrier image for watermarkingH _i ^* Obtaining colour carrier image containing watermarkH ^* Whereini=1, 2, 3 represent red, green, blue three layers, respectively;

the watermark extraction process is described as follows:

the first step: image processing of a water-bearing carrier by a dimension-reducing treatmentH ^* Divided into three layered images of red, green and blueH _i ^* The method comprises the steps of carrying out a first treatment on the surface of the At the same time, each watermark is layered into an imageH _i ^* Is divided into the following sizesm×mNon-overlapping pixel blocks of (1), whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a second step of: layering images sequentially from a watermark in longitudinal orderH _i ^* Is selected to be adjacent to the watermark pixel blockA ^* AndB ^* whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a third step of: directly computing a watermark pixel block in the airspace according to equation (5)A ^* AndB ^* direct current coefficient of (2)dc _p ^* ；

(5)

Wherein,,mis the size of the row (column) of the pixel block,f _p ^* (x, y) Is a water-printing pixel squarepFirst, thexLine 1yThe pixel values of the columns are used to determine,p=1, 2 respectively representing a water-containing print pixel blockA ^* AndB ^* ；

fourth step: according to a water-bearing pixel blockA ^* AndB ^* direct current coefficient of (2)dc _p ^* The size relationship between the pixel blocks is printed from the water by using the formula (6)A ^* AndB ^* extracting watermark bits from a watermarkw*Whereinp=1, 2；

(6)

Fifth step: repeating the second to fourth steps to obtain the extracted binary watermark sequenceSW _i ^* Watermark binary sequenceSW _i ^* Every 8 bits of binary information are divided into a group and converted into decimal pixel values to form a layered watermark image, whereini=1, 2, 3 represent red, green, blue three layers, respectively;

sixth step: key-based hierarchical watermark imageKa _i Is transformed by inverse Arnold to obtain watermark image of each layerW _i ^* The method comprises the steps of carrying out a first treatment on the surface of the Combining watermark extracted images of layersW _i ^* Forming a final extracted watermark imageW ^* Whereini=1, 2, 3 represent three layers of red, green, and blue, respectively.

The method utilizes the correlation rule of the square direct current coefficients of adjacent images to directly obtain the direct current coefficients of discrete cosine transform in a space domain and complete the embedding and blind extraction of the color digital watermark without carrying out real discrete cosine transform; the method not only has better watermark invisibility, but also has stronger watermark robustness and higher algorithm instantaneity.

Drawings

Fig. 1 (a), 1 (b) are two original color carrier images.

Fig. 2 (a) and 2 (b) are two original color watermark images.

Fig. 3 (a) and 3 (b) are watermark images obtained by embedding the watermarks shown in fig. 2 (a) into the carrier images in sequence in fig. 1 (a) and 1 (b), wherein the structural similarity SSIM values are 0.9343 and 0.9351 in sequence, and the peak signal-to-noise ratio PSNR values are 37.4377dB and 37.2156dB in sequence.

Fig. 4 (a) and 4 (b) are watermarks extracted from fig. 3 (a) and 3 (b) in order, and normalized cross-correlation coefficient NC values thereof are 1.0000 and 1.0000, respectively.

Fig. 5 (a), 5 (b), 5 (c), 5 (d), 5 (e), and 5 (f) are watermarks extracted by sequentially subjecting the watermark image shown in fig. 3 (a) to attacks such as JPEG2000 compression (4:1), JPEG compression (70), salt and pepper noise (0.2%), median filtering (3×3), and scaling (75%), and the normalized cross-correlation coefficient NC values thereof are 0.9684, 0.9573, 0.9950, 0.9142, and 0.9626, respectively.

Fig. 6 (a) and 6 (b) show watermark images obtained by embedding the watermarks shown in fig. 2 (b) into the carrier images in sequence in fig. 1 (a) and 1 (b), wherein the structural similarity SSIM values are 0.9346 and 0.9297 in sequence, and the peak signal-to-noise ratio PSNR values are 37.4091dB and 37.2863dB in sequence.

Fig. 7 (a) and 7 (b) are watermarks extracted from fig. 6 (a) and 6 (b) in order, and normalized cross-correlation coefficient NC values thereof are 1.0000 and 1.0000, respectively.

Fig. 8 (a), 8 (b), 8 (c), 8 (d), 8 (e), and 8 (f) are watermarks extracted by sequentially subjecting the watermark image shown in fig. 6 (a) to attacks such as JPEG2000 compression (4:1), JPEG compression (70), salt and pepper noise (0.2%), median filtering (3×3), and scaling (75%), and the normalized cross-correlation coefficient NC values thereof are 0.9676, 0.9526, 0.9956, 0.9018, 0.9673, respectively.

Detailed Description

The invention aims to provide a blind watermarking method of an airspace color digital image fused with discrete cosine transform, which comprises two specific watermarking embedding and watermarking extracting processes, wherein the watermarking embedding process is described as follows:

the first step: a color carrier image with the size of 512 multiplied by 512 is formedHThree layered carrier images separated into red, green and blueH _i And imaging each layered carrierH _i Divided into 2 x 2 non-overlapping pixel blocks, whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a second step of: a color watermark image with the size of 32 multiplied by 32 is displayedWDividing into three layered watermark images of red, green and blue, and simultaneously, carrying out key-based on each layered watermark image in order to improve the security of the watermarkKa _i Arnold transformation to obtain three scrambled layered watermark imagesW _i The method comprises the steps of carrying out a first treatment on the surface of the Layered watermark imageW _i Each decimal pixel value of (a) is converted to an 8-bit binary number (e.g., decimal number 225 may be converted to a binary sequence '11100001'), which in turn is connected to a watermark bit sequenceSW _i ，SW _i Length of 8 x 32 ² =8192, wherei=1, 2, 3 represent red, green, blue three layers, respectively;

and a third step of: sequentially from layered carrier images in longitudinal orderH _i Is selected adjacent pixel square blockAAndBwhereini=1, 2, 3 represent red, green, blue three layers, respectively; here, set upi=1, pixel block selected from red layerAIs that

Selected pixel blockBIs->

；

Fourth step: according to formula (1), directly calculating neighboring pixel blocks in the spatial domainAAndBdirect current coefficient of (2)dc _p ；

(1)

Wherein,,p=1, 2 respectively represents pixel blocksAAndB，mis the size of the row (column) of the pixel block,f _p (x, y) Is a pixel square blockpFirst, thexLine 1yPixel values of the columns; at this time, the size of the row (column) of the pixel blockm=2, calculated pixel blockADirect current coefficient of (2)dc ₁ Pixel block= 448.5000BDirect current coefficient of (2)dc ₂ =448.5000；

Fifth step: sequentially from a layered watermark sequenceSW _i Is selected fromTo be embedded with watermark bitswThe method comprises the steps of carrying out a first treatment on the surface of the Embedding watermark information according to the correlation of DC coefficients of adjacent pixel blocks by using formulas (2) and (3)wObtaining a pixel blockpModified DC coefficientdc _p ^* ，p=1, 2 respectively represents pixel blocksAAndB；

(2)

(3)

wherein,,win order for the watermark bits to be embedded,avg=(dc ₁ +dc ₂ )/2，T _i is the firstiThe quantization step size of the layer is determined,i=1, 2, 3 represent red, green, blue three layers, respectively; at this time, the liquid crystal display device,i=1, from watermark sequenceSW ₁ Selected watermark bits to be embeddedw=‘0’，T ₁ =24.18，avg448.5000, the modified pixel block is obtained according to the formulas (2), (3)A ^* Direct current coefficient of (2)dc ₁ ^* Pixel block= 448.5000B ^* Direct current coefficient of (2)dc ₂ ^* =448.5000；

Sixth step: according to equation (4), using the modified pixel valuef _p ^* (x, y) Replacement of original pixel blockpPixel value of corresponding positionf _p (x, y) Obtaining pixel square block containing watermarkA ^* AndB ^* ；

(4)

wherein the method comprises the steps ofp=1, 2; at this time, the obtained water-containing print pixel blockA ^* Is that

Water-printing pixel squareB ^* Is->

；

Seventh step: print pixel blockA ^* AndB ^* respectively updated to its layered carrier imageH _i Corresponding positions in (a), whereini=1, 2, 3, respectively representing three layers of red, green, blue; at this time, the liquid crystal display device,i=1, watered pixel blockA ^* AndB ^* is updated to its layered carrier image respectivelyH ₁ Corresponding positions in (a);

eighth step: repeating the third to seventh steps until all watermark information is embedded, thereby obtaining layered carrier image containing watermarkH _i ^* The method comprises the steps of carrying out a first treatment on the surface of the Finally, combining three layered watermark-containing layered carrier imagesH _i ^* Obtaining colour carrier image containing watermarkH ^* Whereini=1, 2, 3 represent red, green, blue three layers, respectively;

the watermark extraction process is described as follows:

the first step: image processing of a water-bearing carrier by a dimension-reducing treatmentH ^* Divided into three layered images of red, green and blueH _i ^* The method comprises the steps of carrying out a first treatment on the surface of the At the same time, each watermark is layered into an imageH _i ^* Divided into non-overlapping pixel blocks of size 2 x 2, whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a second step of: layering images sequentially from a watermark in longitudinal orderH _i ^* Is selected to be adjacent to the watermark pixel blockA ^* AndB ^* whereini=1, 2, 3 represent red, green, blue three layers, respectively; at this time, set upi=1, a block of water-containing pixels selected from the red layerA ^* Is that

Pixel block containing watermarkB ^* Is->

；

And a third step of: direct computation of watermark pixel blocks in airspace according to equation (5)A ^* AndB ^* direct current coefficient of (2)dc _p ^* ；

(5)

Wherein,,mis the size of the row (column) of the pixel block,f _p ^* (x, y) Is a water-printing pixel squarepFirst, thexLine 1yThe pixel values of the columns are used to determine,p=1, 2 respectively representing a water-containing print pixel blockA ^* AndB ^* the method comprises the steps of carrying out a first treatment on the surface of the At this time, the size of the row (column) of the pixel blockm=2, calculated pixel blockA ^* Direct current coefficient of (2)dc ₁ ^* Pixel block= 448.5000B ^* Direct current coefficient of (2)dc ₂ ^* =448.5000；

Fourth step: according to a water-bearing pixel blockA ^* AndB ^* direct current coefficient of (2)dc _p ^* The size relationship between the pixel blocks is printed from the water by using the formula (6)A ^* AndB ^* extracting watermark bits from a watermarkw ^* Whereinp=1, 2；

(6)

At this time, the liquid crystal display device,dc ₁ ^* -dc ₂ ^* =0, the extracted watermark bits are obtained according to equation (6)w ^* =‘0’；

Fifth step: repeatedly executing the second to the first steps of the processFour steps, obtaining the extracted binary watermark sequenceSW _i ^* Watermark binary sequenceSW _i ^* Every 8 bits of binary information is divided into a group and converted into decimal pixel values to form a layered watermark image, whereini=1, 2, 3 represent red, green, blue three layers, respectively;

sixth step: key-based hierarchical watermark imageKa _i Is transformed by inverse Arnold to obtain watermark image of each layerW _i ^* The method comprises the steps of carrying out a first treatment on the surface of the At the same time, the watermark extraction images of the layers are combinedW _i ^* Forming a final extracted watermark imageW ^* Whereini=1, 2, 3 represent three layers of red, green, and blue, respectively.

The method has better watermark invisibility, stronger robustness and higher algorithm instantaneity, and is suitable for copyright protection of color digital images as watermarks.

The invention has the effect of verification

To prove the effectiveness of the present invention, two standard 24-bit color images of 512×512 size as shown in fig. 1 (a) and 1 (b) were selected as carrier images, and two 24-bit color images of 32×32 size as shown in fig. 2 (a) and 2 (b) were used as digital watermarks, respectively, for verification.

Fig. 3 (a) and 3 (b) are watermark images obtained by embedding the watermarks shown in fig. 2 (a) into the carrier images in sequence in fig. 1 (a) and 1 (b), wherein the structural similarity SSIM values are 0.9343 and 0.9351 in sequence, and the peak signal-to-noise ratio PSNR values are 37.4377dB and 37.2156dB in sequence; fig. 4 (a) and 4 (b) are watermarks extracted from fig. 3 (a) and 3 (b) in sequence, and normalized cross-correlation coefficient NC values thereof are 1.0000 and 1.0000, respectively; fig. 5 (a), 5 (b), 5 (c), 5 (d), and 5 (e) are watermarks extracted by sequentially subjecting the watermark image shown in fig. 3 (a) to attacks such as JPEG2000 compression (4:1), JPEG compression (70), salt and pepper noise (0.2%), median filtering (3×3), and scaling (75%), and the normalized cross-correlation coefficient NC values thereof are 0.9684, 0.9573, 0.9950, 0.9142, 0.9626, respectively.

Fig. 6 (a) and 6 (b) are watermark images obtained by embedding the watermarks shown in fig. 2 (b) into the carrier images in sequence in fig. 1 (a) and 1 (b), wherein the structural similarity SSIM values are 0.9346 and 0.9297 in sequence, and the peak signal-to-noise ratio PSNR values are 37.4091dB and 37.2863dB in sequence; fig. 7 (a) and 7 (b) are watermarks extracted from fig. 6 (a) and 6 (b) in order, and normalized cross-correlation coefficient NC values thereof are 1.0000 and 1.0000, respectively; fig. 8 (a), 8 (b), 8 (c), 8 (d), 8 (e), and 8 (f) are watermarks extracted after attacks such as JPEG2000 compression (4:1), JPEG compression (70), salt and pepper noise (0.2%), median filtering (3×3), and scaling (4:1) are sequentially performed on the watermark image shown in fig. 6 (a), and normalized cross-correlation coefficient NC values thereof are 0.9676, 0.9526, 0.9956, 0.9018, 0.9673, respectively.

The algorithm runs on the platform 2.60GHZ CPU,4.00GB RAM,Win10,MATLAB (R2017 a) for nearly ten thousand times, the average embedding time of the digital watermark is 0.605655 seconds, the average extraction time is 0.255208 seconds, and the total time is 0.860863 seconds.

In summary, the embedded digital image watermark has higher invisibility, and meets the invisibility requirement of a watermark algorithm; meanwhile, the digital image watermark extracted from various attacked images has better authenticability and higher NC value, which indicates that the method has stronger robustness; in addition, the average running total time of the algorithm is less than 1 second, and the requirement of rapid copyright protection of multimedia big data is met.

Claims (1)

1. A blind watermarking method of a space domain color digital image fused with discrete cosine transform is characterized in that: the method is realized through a specific watermark embedding process and a watermark extraction process, and the watermark embedding process is described as follows:

the first step: one width is of the sizeM×MIs a color carrier image of (a)HThree layered carrier images separated into red, green and blueH _i And imaging each layered carrierH _i Is divided intom×mNon-overlapping pixel blocks of (1), whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a second step of: one width is of the sizeN×NColor watermark image of (a)WDividing into three layered watermark images of red, green and blue, and simultaneously, carrying out key-based on each layered watermark image in order to improve the security of the watermarkKa _i Arnold transformation to obtain three scrambled layered watermark imagesW _i The method comprises the steps of carrying out a first treatment on the surface of the Layered watermark imageW _i Each decimal pixel value in (2) is converted into 8-bit binary number which is sequentially connected into a length of 8N ² Watermark bit sequence of (a)SW _i Whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a third step of: sequentially from layered carrier images in longitudinal orderH _i Is selected adjacent pixel square blockAAndBwhereini=1, 2, 3 represent red, green, blue three layers, respectively;

fourth step: according to formula (1), pixel blocks are directly calculated in the space domainAAndBdirect current coefficient of (2)dc _p ；

(1)

Wherein,,p=1, 2 respectively represents pixel blocksAAndB，mis the size of the row (column) of the pixel block,f _p (x, y) Is a pixel square blockpFirst, thexLine 1yPixel values of the columns;

fifth step: sequentially from a layered watermark sequenceSW _i Selecting watermark bits to be embedded inwThe method comprises the steps of carrying out a first treatment on the surface of the Embedding watermark information according to the correlation of DC coefficients of adjacent pixel blocks by using formulas (2) and (3)wObtaining a pixel blockpModified DC coefficientdc _p ^* ，p=1, 2 respectively represents pixel blocksAAndB；

(2)

(3)

wherein,,win order for the watermark bits to be embedded,avg=(dc ₁ +dc ₂ )/2，T _i is the firstiThe quantization step size of the layer is determined,i=1, 2, 3 represent red, green, blue three layers, respectively;

sixth step: according to equation (4), using the modified pixel valuef _p ^* (x, y) Replacement of original pixel blockpPixel value of corresponding positionf _p (x, y) Obtaining pixel square block containing watermarkA ^* AndB ^* whereinp=1, 2；

(4)

Seventh step: print pixel blockA ^* AndB ^* respectively updated to its layered carrier imageH _i Corresponding positions in (a), whereini=1, 2, 3, respectively representing three layers of red, green, blue;

eighth step: repeating the third to seventh steps until all watermark information is embedded, thereby obtaining layered carrier image containing watermarkH _i ^* The method comprises the steps of carrying out a first treatment on the surface of the Finally, combining three layered watermark-containing layered carrier imagesH _i ^* Obtaining colour carrier image containing watermarkH ^* Whereini=1, 2, 3 represent red, green, blue three layers, respectively;

the watermark extraction process is described as follows:

the first step: image processing of a water-bearing carrier by a dimension-reducing treatmentH ^* Divided into three layered images of red, green and blueH _i ^* The method comprises the steps of carrying out a first treatment on the surface of the At the same time, each watermark is layered into an imageH _i ^* Is divided into the following sizesm×mNon-overlapping pixel blocks of (1), whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a second step of: layering images sequentially from a watermark in longitudinal orderH _i ^* Is selected to be adjacent to the watermark pixel blockA ^* AndB ^* whereini=1, 2, 3 represent red, green, blue three layers, respectively;

and a third step of: directly computing a watermark pixel block in the airspace according to equation (5)A ^* AndB ^* direct current coefficient of (2)dc _p ^* ；

(5)

Wherein,,mis the size of the row (column) of the pixel block,f _p ^* (x, y) Is a water-printing pixel squarepFirst, thexLine 1yThe pixel values of the columns are used to determine,p=1, 2 respectively representing a water-containing print pixel blockA ^* AndB ^* ；

fourth step: according to a water-bearing pixel blockA ^* AndB ^* direct current coefficient of (2)dc _p ^* The size relationship between the pixel blocks is printed from the water by using the formula (6)A ^* AndB ^* extracting watermark bits from a watermarkw*Whereinp=1, 2；

(6)

Fifth step: repeating the second to fourth steps to obtain the extracted binary watermark sequenceSW _i ^* Watermark binary sequenceSW _i ^* Every 8 bits of binary information is divided into groups and converted intoDecimal pixel values forming a layered watermark image, whereini=1, 2, 3 represent red, green, blue three layers, respectively;

sixth step: key-based hierarchical watermark imageKa _i Is transformed by inverse Arnold to obtain watermark image of each layerW _i ^* The method comprises the steps of carrying out a first treatment on the surface of the Combining watermark extracted images of layersW _i ^* Forming a final extracted watermark imageW ^* Whereini=1, 2, 3 represent three layers of red, green, and blue, respectively.

Images